How do we identify and measure risk in human–AI interaction?

This breakout session from IASEAI 2026 explores how AI systems behave in real-world contexts—focusing on alignment, privacy trade-offs, personalization, and safety testing across both digital and physical environments.

Across two segments, speakers examine how large language models and autonomous systems can be evaluated, stress-tested, and governed—highlighting the challenges of detecting failure modes, safeguarding users, and designing more reliable AI systems.

Florian Mai — AI Alignment Strategies from a Risk Perspective: Independent Alignment Mechanisms or Shared Failures?
Soumi Das — Revisiting Privacy, Utility, and Efficiency Trade-offs when Fine-Tuning Large Language Models
Abhisek Dash — The Algorithmic Self-Portrait: Deconstructing Memory in ChatGPT
Nathan Henry — Replt: Steering Language Models with Concept-Specific Refusal Vectors
Shirali Nigam — Crash-Test Dummies for AI: Detecting Mental Health Risks to Safeguard Users
Gen Li — Gen-NCAP: A Generative Simulator for Corner Case Benchmarking in End-to-End Autonomous Driving
Hao Zhao — Challenger: Affordable Adversarial Driving Video Generation for Safety Testing

📍 Recorded at UNESCO Headquarters, Paris

The International Association for Safe and Ethical AI (IASEAI) is a global professional association bringing together researchers, policymakers, and practitioners working to ensure that advanced AI systems operate safely and ethically for the benefit of humanity.

Learn more: https://www.iaseai.org

#AI #AISafety #AIGovernance #AIpolicy #IASEAI #ResponsibleAI